THE LAMBDA COMPLEX: KNOWING YOUR PLACE IN THE THRESHOLD MATRIX

Purpose: Within a budget constrained health care system, the cost-effectiveness threshold is typically assumed to represent the incremental cost-effectiveness ratio (ICER) of the existing technology displaced as a result of investing additional resources in a new technology. There is ongoing debate as to whether the threshold should take account of different assumptions regarding the decision maker’s objective, the marginality of the new technology, and other considerations. This paper attempts to inform this debate by comprehensively simulating a wide range of alternative assumptions, allowing for consideration of the impact of each assumption on the threshold.

Method: A model of a hypothetical health care system was constructed using Microsoft Excel 2013. A large number of health technologies were modeled, and the budget of the health system was fixed such that only some technologies could be adopted at any one time. The model simulated the initial allocation of technologies, which may or may not be allocatively efficient. It then simulated displacement resulting from investment in a new technology, or further adoption of technologies made possible as a result of disinvestment, in order to determine the appropriate threshold to adopt for investment or disinvestment decisions respectively. The simulation was repeated for each combination of a number of possible assumptions regarding: the decision maker’s objective; the budget impact of the investment or disinvestment (marginal or non-marginal); the decision maker’s authority; political considerations; marginal returns (constant or diminishing); the information available to the decision maker (perfect or imperfect); and several other considerations. The appropriate thresholds were aggregated across all simulations to create a multi-dimensional “threshold matrix”, allowing the impact of each assumption to be assessed in each possible context.

Result: Each assumption had the potential to influence the threshold, with the impact dependent upon other assumptions adopted. In many cases, including (but not limited to) where the initial allocation of technologies is allocatively inefficient, or where the decision maker faces imperfect information and/or political considerations, the appropriate threshold for investment decisions differs from that for disinvestment decisions.

Conclusion: The “threshold matrix” allows for consideration of the impact of alternative assumptions on the appropriate thresholds to adopt. In real world health care systems – subject to inefficiencies, imperfect information and political considerations – different thresholds ought to be adopted for investment and disinvestment decisions.